Chrysin enhances anticancer drug-induced toxicity mediated by the reduction of claudin-1 and 11 expression in a spheroid culture model of lung squamous cell carcinoma cells

The aberrant expression of claudins (CLDNs), which are tight junctional proteins, is seen in various solid tumors, but the regulatory mechanisms and their pathophysiological role are not well understood. Both CLDN1 and CLDN11 were highly expressed in human lung squamous cell carcinoma (SCC). Chrysin, found in high concentration in honey and propolis, decreased CLDN1 and CLDN11 expression in RERF-LC-AI cells derived from human lung SCC. The phosphorylation level of Akt was decreased by chrysin, but those of ERK1/2 and c-Jun were not. LY-294002, an inhibitor of phosphatidylinositol 3-kinase, inhibited the phosphorylation of Akt and decreased the expression levels of CLDN1 and CLDN11. The association between phosphoinositide-dependent kinase 1 (PDK1) and Akt was inhibited by chrysin, but the phosphorylation of PDK1 was not. Immunoprecipitation and quartz-crystal microbalance assays revealed that biotinylated-chrysin binds directly to Akt. The knockdown of CLDN1 and CLDN11 using small interfering RNAs increased the transepithelial flux of doxorubicin (DXR), an anthracycline anticancer drug. Similarly, both chrysin and LY-294002 increased DXR flux. Neither CLDN1 knockdown, CLDN11 knockdown, nor chrysin changed the anticancer drug-induced cytotoxicity in a two-dimensional culture model, whereas they enhanced cytotoxicity in a spheroid culture model. Taken together, chrysin may bind to Akt and inhibit its phosphorylation, resulting in the elevation of anticancer drug-induced toxicity mediated by reductions in CLDN1 and CLDN11 expression in RERF-LC-AI cells. We suggest that chrysin may be useful as an adjuvant chemotherapy in lung SCC.

Effects of knockdown of CLDN1 and CLDN11 on sensitivity to anticancer drugs. The pathophysiological roles of CLDN1 and CLDN11 have not been fully clarified in lung SCC cells. We examined the effects of CLDN1 and CLDN11 knockdown on anticancer agent-induced cytotoxicity. The introduction of small interfering RNAs (siRNAs) for CLDN1 and CLDN11 suppressed the protein expression of CLDN1 and CLDN11, respectively ( Fig. 2A). DXR and CDDP increased cytotoxicity in a dose-dependent manner (Fig. 2B). Neither CLDN1 nor CLDN11 knockdown significantly changed the sensitivity to DXR and CDDP. Anticancer-induced cytotoxicity was diminished by the activation of their extrusion through ABC transporters 22 . Neither CLDN1 nor CLDN11 knockdown significantly changed the protein levels of ABCB1, ABCC1, or ABCG2 (Fig. 2C). The expression of ABCC2 was detected in positive control A549 cells, but not in RERF-LC-AI cells. These results indicated that neither CLDN1 nor CLDN11 is involved in chemosensitivity in this 2D culture model.

Effects of sodium caprate (CA) and knockdown of CLDN on transepithelial permeability. CLDN
subtypes can form homo-or heterophilic interactions between adjacent cells and regulate paracellular solute and ion transport [23][24][25] . Transepithelial electrical resistance (TER) was increased after days 2 and made a plateau phase (Fig. 3A). TER was significantly decreased by CA, a TJs modulator (Fig. 3B). In addition, transepithelial fluxes of lucifer yellow, a fluorescent marker for the paracellular pathway, and DXR were increased by CA (Fig. 3C). Both TER and transepithelial fluxes of lucifer yellow were changed by CA, but RERF-LC-AI cells did not form continuous TJs and their TJ barrier was incomplete. Both CLDN1 and CLDN11 knockdown increased transepithelial permeability to DXR without changing TER (Fig. 3D,E). These results indicated that CLDN1 and CLDN11 may not be involved in the regulation of paracellular ion permeability, but they may suppress paracellular permeability for small molecules.
Increase in DXR sensitivity by knockdown of CLDN1 and CLDN11 in a 3D spheroid model. Neither CLDN1 nor CLDN11 changed spheroid sizes, but they significantly decreased hypoxia levels ( Fig. 4A,B). The fluorescence intensity of DXR in the spheroids was increased in a dose dependent manner, indicating that DXR accumulated in the spheroids (Fig. 4C). The accumulation of DXR was significantly enhanced by CLDN1 and CLDN11 knockdown. DXR decreased the viability of spheroid cells in a dose-dependent manner, which was enhanced by CLDN1 and CLDN11 knockdown (Fig. 4D). These results indicated that both CLDN1 and CLDN11 may be involved in chemoresistance in 3D spheroid cells.

Involvement of Akt in the regulation of CLDN1 expression.
There are no reports on what intracellular signaling factors are involved in CLDN1 and CLDN11 expression in lung SCC cells. Chrysin did not change p-ERK1/2 or p-c-Jun levels, whereas it decreased p-Akt levels (Fig. 6A). The phosphorylation of Akt is upregulated by phosphoinositide-dependent kinase 1 (PDK1), a kinase functioning downstream of phosphatidylinositol 3-kinase (PI3K). However, p-PDK1 levels were not changed by chrysin. The total amounts of Akt, PDK1, ERK1/2, and c-Jun were not changed by chrysin. LY-294002, an inhibitor of the PI3K/Akt signaling pathway, significantly decreased p-Akt, CLDN1, and CLDN11 levels (Fig. 6B,C). In addition, the mRNA levels of CLDN1 and CLDN11 were decreased by LY-294002 (Fig. 6D). These results coincided with those using chrysin.

Inhibition of the association between PDK1 and Akt by chrysin.
To clarify the direct interaction of chrysin, we synthesized biotinylated-chrysin (Fig. 7A). In immunoprecipitation assays, biotinylated-chrysin bound to Akt, but not to PDK1 (Fig. 7B). In addition, QCM analysis showed that biotinylated-chrysin bound to  www.nature.com/scientificreports www.nature.com/scientificreports/ Effects of chrysin on sensitivity and transepithelial permeability to anticancer drugs. As shown in Fig. 5A, chrysin did not show cytotoxicity at 10 μM. Chrysin slightly suppressed the DXR-induced cytotoxicity at 3-10 μM, whereas it did not affect CDDP-induced cytotoxicity (Fig. 8A). Next, we examined the effects of chrysin and LY-294002 on transepithelial permeability. Both chrysin and LY-294002 increased transepithelial permeability to DXR without changing TER (Fig. 8B). These results coincided with those using CLDN1 and CLDN11 knockdown.

Increase in DXR and CDDP toxicities of spheroid cells by chrysin.
Chrysin did not change spheroid size, but it significantly decreased hypoxia levels ( Fig. 9A,B). The accumulation of DXR in the spheroids was significantly enhanced by chrysin (Fig. 9C). DXR and CDDP decreased the viability of spheroid cells in a dose-dependent manner, which was enhanced by chrysin ( Fig. 9D,E). These results coincided with those using CLDN1 and CLDN11 knockdown. Chrysin may enhance the sensitivity to anticancer drugs in 3D spheroid cells mediated by the reduction in CLDN1 and CLDN11 expression. To support the idea, we examined the effect of CLDN1 overexpression on the accumulation and toxicity of DXR. The chrysin-induced elevation of accumulation and toxicity of DXR was significantly inhibited by CLDN1 overexpression (Fig. 10).

Discussion
Normal lung epithelia expresses CLDN1, 3, 4, 5, 7, and 18, but not CLDN11 27,28 . We reported previously that the expression of CLDN1 is upregulated in human SCC tissue and RERF-LC-AI cells compared with normal tissue, whereas CLDN3, 4, 5, 7, and 18 are downregulated 19 . Here, we found that CLDN11 is highly expressed in SCC ( Fig. 1). Immunofluorescence and TER measurements showed the cells form incomplete TJs. However, transepithelial fluxes of lucifer yellow and DXR were increased by the treatment with CA or introduction of siRNA for CLDN1 and CLDN11, suggesting that the cells can form partially functional TJs barrier. The elevation of CLDN1 expression was reported in various carcinoma tissues including colon 29 , stomach 30 , and pancreas 31 . Similar to our results, Paschoud et al. 32 reported that the mRNA and protein levels of CLDN1 in SCC are higher than those in parenchyma. However, their report indicated no correlation between the expression of CLDN1 and the clinicopathological characteristics of SCC.
The downregulation of CLDN3 and CLDN7 is associated with poor prognosis and poor survival in patients with SCC, respectively 33,34 . The overexpression of CLDN5, 7, and 18 suppresses cell cycle progression from G1 to S phase in RERF-LC-AI cells, indicating that these CLDNs are involved in the regulation of SCC cell proliferation 19 . In contrast, the pathophysiological roles of CLDN1 and CLDN11 in SCC are not fully understood. We reported recently that the expression of CLDN1 is exaggerated by CDDP resistance in A549 cells 10 . The www.nature.com/scientificreports www.nature.com/scientificreports/ knockdown of CLDN1 and CLDN11 expression by siRNA increased the transepithelial flux of DXR, but it did not significantly change anticancer drug-induced toxicity or the expression of ABC transporters in a 2D culture model of RERF-LC-AI cells (Figs 2 and 3), suggesting that neither CLDN1 nor CLDN11 are directly involved in chemoresistance against anticancer drugs.
The PI3K/Akt signaling pathway is mainly involved in the regulation of cell survival and apoptosis 35 , and the activation of Akt has been reported in over 60% of NSCLC patients 36 . Both CLDN1 and CLDN11 expression levels were decreased by LY-294002 (Fig. 6), suggesting that their expression is upregulated by Akt in SCC cells. Similarly, the expression of CLDN1 is upregulated by Akt in A549 19 and cervical adenocarcinoma cells 37 . In contrast, an inverse relationship between phosphorylation of Akt and CLDN1 expression is reported in esophageal SCC cells 38 . The regulatory mechanism of CLDN1 expression may differ according to the type of carcinoma. On the other hand, there are no reports showing whether Akt is involved in the regulation of CLDN11 expression. Notably, chrysin decreased both CLDN1 and CLDN11 expression (Fig. 5). Furthermore, chrysin significantly decreased p-Akt levels, but did not change p-ERK1/2 and p-c-Jun levels. We suggest that chrysin decreases both CLDN1 and CLDN11 in RERF-LC-AI cells mediated by the inhibition of p-Akt.
Chrysin shows a preventive effect on cancer mediated by various mechanisms including reduction in the activities of cytochrome P450-dependent monooxygenases, induction of the activity of antioxidant and detoxification enzymes, inhibition of cellular proliferation, and induction of apoptosis 39 . The phenotype of cancer cells is regulated by diverse signaling molecules including the Akt pathway. Chrysin decreased p-Akt levels without affecting the total amount of Akt or phosphorylation of PDK1 (Fig. 6A). The inhibition of Akt by chrysin has been reported in A549 12 , breast cancer 40 and leukemia cells 41 . Therefore, it is suggested that chrysin binds to Akt, although there was no direct evidence showing this as yet. Immunoprecipitation and QCM assays revealed that biotinylated chrysin can bind directly to Akt (Fig. 7B,C). The association between PDK1 and Akt was inhibited by chrysin (Fig. 7D). Thus, our data provide the first indication that chrysin can interact directly with Akt, resulting in inhibition of the phosphorylation of Akt and association of PDK1 with Akt.
The knockdown of CLDN1 and CLDN11 by siRNA enhances DXR accumulation and DXR-induced toxicity in 3D spheroid cells (Fig. 4). Similar results were observed using treatment with chrysin (Fig. 9B). Transepithelial flux of DXR was negatively regulated by CLDN1 and CLDN11. The accumulation and toxicity of DXR was www.nature.com/scientificreports www.nature.com/scientificreports/ significantly suppressed by CLDN1 overexpression (Fig. 10). We suggest that CLDN1 or CLDN11 overexpression impaired tumor sensitivity to anticancer drugs mediated by interference with penetration of the drugs into inner areas of the spheroid. Another explanation is that both CLDN1 and CLDN11 induce chemoresistant characteristics of cancer cells. Hypoxia levels of spheroid were inversely changed by CLDN1 and CLDN11 expression (Fig. 4). Cancer cells form microenvironment in vivo, and the inner cells of the microenvironment are exposed to nonlethal hypoxia and oxidative stress, which has significant effects on tumor progression and treatment efficacy. These findings raise a possibility that the alleviation of hypoxic conditions by decreasing in the expression of CLDN1 and CLDN11 improves chemosensitivity in spheroid cells. High interstitial fluid pressure, increased collagen production, and poorly formed vasculature exacerbate hypoxia 42 . However, it is unknown what mechanisms are involved in the establishment of hypoxic conditions in vitro in spheroids. Further studies are needed to clarify how CLDN and chrysin improve them, but chrysin may be useful to improve hypoxia and suppress the malignancy of SCC cells.
In conclusion, we found that human SCC tissue and RERF-LC-AI cells exhibit high expression levels of not only CLDN1 but also CLDN11 compared with normal tissue. Chrysin inhibited the phosphorylation of Akt and decreased the expression levels of CLDN1 and CLDN11 similar to LY-294002. Immunoprecipitation and QCM www.nature.com/scientificreports www.nature.com/scientificreports/ assays showed that chrysin binds directly to Akt and inhibits the association of PDK1 with Akt. Chrysin increased transepithelial flux of DXR without affecting TER. In addition, chrysin did not change anticancer agent-induced toxicity in a 2D model, but it enhanced toxicity in a 3D spheroid model. Our data indicate that chrysin may be a potential compound for adjuvant treatment of human SCC.

Material and Methods
Materials. Antibodies used in the present experiments were listed in Table 1 2D and 3D Cell culture. RERF-LC-AI cells (RIKEN BRC through the National Bio-Resource Project of the MEXT, Ibaraki, Japan) were cultured as described previously 10 . For 3D culture, the cells were plated at densities of 1 × 10 4 cells/well on PrimeSurface96V multi-well plates (Sumitomo Bakelite, Tokyo, Japan). After culturing for 96 h, the size and viability of spheroids were measured as described previously 10 . The fluorescence intensities of DXR and LOX-1, a hypoxia probe, were calculated using ImageJ software. siRnA and transfection. siRNAs for negative control and CLDNs were obtained from Santa Cruz and Sigma-Aldrich, respectively. The siRNAs were transfected into 2D and 3D cultured cells using Lipofectamine 2000 and ScreenFect A, respectively. www.nature.com/scientificreports www.nature.com/scientificreports/ using Lung Cancer cDNA Array II and V (OriGene, Rockville, MD, USA). Reverse transcription was carried out using a ReverTra Ace qPCR RT Kit (Toyobo Life Science, Osaka, Japan). Quantitative real-time PCR was performed using an Eco Real-Time polymerase chain reaction (PCR) system (AS One, Osaka, Japan) with a THUNDERBIRD SYBR qPCR Mix (Toyobo Life Science). The primers used for PCR are listed in Table 2. The relative change in mRNA expression was calculated as described previously 10 .
Sodium dodecyl sulfate-polyacrylamide gel electrophoresis and western blotting. Confluent cells were scraped into cold phosphate-buffered saline and precipitated by centrifugation. Then, the cells were lysed in a RIPA buffer (150 mM NaCl, 50 mM Tris-HCl (pH 8.0), 1% Triton X-100, 0.1% sodium dodecyl sulfate (SDS), 0.5 mM EDTA) supplemented with a protease inhibitor cocktail (Sigma-Aldrich), and sonicated for 20 s. Nuclear fraction was removed by centrifugation at 6,000 × g for 5 min. The resultant supernatants were used as cell lysates. SDS-polyacrylamide gel electrophoresis and western blotting were carried out as described previously 10 . www.nature.com/scientificreports www.nature.com/scientificreports/ immunoprecipitation. Cell lysates were incubated with Protein G-sepharose (GE Healthcare, Bucks, UK) and anti-PDK1 antibodies for 16 h at 4 °C using a rotator. In the case using biotinylated-chrysin, cell lysates were incubated with or without biotinylated-chrysin in the presence of avidin agarose (Thermo Fisher Scientific). Immunoprecipitants were washed four times with an immunoprecipitation buffer (150 mM NaCl, 0.5 mM EDTA, 0.5% Triton X-100, 50 mM Tris-HCl (pH 7.4), and a protease inhibitor cocktail)and then subjected to SDS-polyacrylamide gel electrophoresis. confocal microscopy. Immunofluorescence measurements were carried out as described previously 10 .
Measurement of transepithelial permeability. Cells were seeded at densities of 5 × 10 4 cells on Transwells (0.4 μm pore size, 12 mm diameter) with polyester membrane inserts (Corning Incorporated, Corning, NY, USA). TER and paracellular flux of lucifer yellow and DXR were measured as describe previously 10 .